From: Alex Sadleir Date: Sun, 14 Apr 2013 06:48:31 +0000 Subject: enforce valid package url X-Git-Url: https://maxious.lambdacomplex.org/git/?p=disclosr.git&a=commitdiff&h=4bfca5ebcf1771fdd497046e81de89596f6f640d --- enforce valid package url Former-commit-id: 26accf46eff2958223ffae9a1234d4291379bce4 --- --- a/documents/datagov-export.py +++ b/documents/datagov-export.py @@ -91,9 +91,10 @@ name = re.sub('__', '_', name).lower() return name - +#todo "{'name': [u'Url must be purely lowercase alphanumeric (ascii) characters and these symbols: -_']}" +# http://data.gov.au/dataset/australian-domestic-regional-and-international-airline-activity-%E2%80%93-time-series/ def name_munge(input_name): - return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and')) + return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and')) #[:100] #return input_name.replace(' ', '').replace('.', '_').replace('&', 'and') @@ -124,7 +125,7 @@ print doc.id if doc.value['url'] != "http://data.gov.au/data/" and doc.value['agencyID'] != "qld": # Collect the package metadata. - pkg_name = doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/','')[:100]; + pkg_name = filter( lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/','')[:100]); tags = [] if doc.value['agencyID'] == "AGIMO": if len(doc.value['metadata']["Keywords / Tags"]) > 0: @@ -166,71 +167,73 @@ #add to group - # group_name = name_munge(doc.value['metadata']["Agency"][:100]) + group_name = name_munge(doc.value['metadata']["Agency"][:100]) + try: + print ckan.group_entity_get(group_name) + + # Update the group details + group_entity = ckan.last_message + print "group exists" + if 'packages' in group_entity.keys(): + group_entity['packages'] = list(set(group_entity['packages'] + [pkg_name])) + else: + group_entity['packages'] = [pkg_name] + ckan.group_entity_put(group_entity) + except CkanApiError, e: + if ckan.last_status == 404: + print "group does not exist, creating" + group_entity = { + 'name': group_name, + 'title': doc.value['metadata']["Agency"], + 'description': doc.value['metadata']["Agency"], + 'packages': [pkg_name], + # 'type': "organization" # not allowed via API, use database query + # update "group" set type = 'organization'; + } + print group_entity + ckan.group_register_post(group_entity) + elif ckan.last_status == 409: + print "group already exists" + else: + raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % ( + ckan.last_status, pkg_name, e.args)) + + # #add to or create organization using direct API + # org_name = name_munge(doc.value['metadata']["Agency"][:100]) + # if org_name not in orgs_list: + # orgs_list = ckandirect.action.organization_list()['result'] + # print orgs_list + # if org_name not in orgs_list: + # try: + # print "org not found, creating "+org_name + # ckandirect.action.organization_create(name = org_name, title= doc.value['metadata']["Agency"], + # description= doc.value['metadata']["Agency"]) + # orgs_list.append(org_name) + # except ckanapi.ValidationError, e: + # print e + # raise LoaderError('Unexpected status') + # # try: - # print ckan.group_entity_get(group_name) - # - # # Update the group details - # group_entity = ckan.last_message - # print "group exists" - # if 'packages' in group_entity.keys(): - # group_entity['packages'] = list(set(group_entity['packages'] + [pkg_name])) - # else: - # group_entity['packages'] = [pkg_name] - # ckan.group_entity_put(group_entity) - # except CkanApiError, e: - # if ckan.last_status == 404: - # print "group does not exist, creating" - # group_entity = { - # 'name': group_name, - # 'title': doc.value['metadata']["Agency"], - # 'description': doc.value['metadata']["Agency"], - # 'packages': [pkg_name], - # # 'type': "organization" # not allowed via API, use database query - # # update "group" set type = 'organization'; - # } - # print group_entity - # ckan.group_register_post(group_entity) - # elif ckan.last_status == 409: - # print "group already exists" - # else: - # raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % ( - # ckan.last_status, pkg_name, e.args)) - - #add to or create organization using direct API - org_name = name_munge(doc.value['metadata']["Agency"][:100]) - if org_name not in orgs_list: - orgs_list = ckandirect.action.organization_list()['result'] - print orgs_list - if org_name not in orgs_list: - try: - print "org not found, creating "+org_name - ckandirect.action.organization_create(name = org_name, title= doc.value['metadata']["Agency"], - description= doc.value['metadata']["Agency"]) - orgs_list.append(org_name) - except ckanapi.ValidationError, e: - print e - raise LoaderError('Unexpected status') - - try: - print "linking package to "+org_name - - #brute force - org = ckandirect.action.organization_show(id=org_name) - #print org["result"]["packages"] - #print pkg - org["result"]["packages"].append({"id":pkg['id'], "name": pkg_name}) - #print org["result"] - ckandirect.call_action(action="organization_update", data_dict=org["result"]) - - # if only this worked - #print ckandirect.action.package_owner_org_update(id=pkg['id'],organization_id=org["result"]["id"]) - - except ckanapi.ValidationError, e: - print e - raise LoaderError('Unexpected status') - #except ckanapi.CKANAPIError, e: - # print "lol" + # print "linking package to "+org_name + # + # #brute force + # org = ckandirect.action.organization_show(id=org_name) + # #print org["result"]["packages"] + # #print pkg + # org["result"]["packages"].append({"id":pkg['id'], "name": pkg_name}) + # #print org["result"] + # ckandirect.call_action(action="organization_update", data_dict=org["result"]) + # + # # if only this worked + # print ckandirect.action.package_owner_org_update(id=pkg['id'],organization_id=org["result"]["id"]) + # + # except ckanapi.ValidationError, e: + # print e + # raise LoaderError('Unexpected status') + # except ckanapi.CKANAPIError, e: + # print "lol" + + if 'Download' in doc.value['metadata'].keys(): try: